CLARK, TANNENBALM
Studies on Limit-Peptide Pigments from Glucose-Casein Browning Systems Using Radioactive Glucose Allen V. Clark1 and Steven R. Tannenbaum*
Protein-bound brown pigments were isolated from glucose-casein . reaction systems by the method of Clark and Tannenbaum (1970). Investigations with 2-14C-, 6-14C-, U-14C-, and 6-Tlabeled glucose showed high correlation between
Numerous investigators have used tracer techniques to study the browning reaction. For example, the origin of COz has been shown to be mostly derived from decarboxylation of amino acids (Chichester et al., 1952). The fate of the carbon skeleton of glucose has also been investigated (Stadtman et al., 1952; Wolfrom et al., 1953), and radioactive nondialyzable pigments have been isolated from a glucose-glycine browning system. In a previous paper (Clark and Tannenbaum, 1970) we presented a general method for isolation and purification of brown limit-peptide pigments from protein-aldose browning systems. This paper presents data on incorporation of glucose labeled with both tritium and 14C into these pigments and thereby extends our present knowledge of pigment structure.
6-T loss from incorporated sugar residues and the amount of brown color. A high molecular weight pigment was isolated in which the glucose-derived moiety lost 26% of tritium bound to C-6.
of each isotope and their ratios (C/T) was determined by appropriate internal standards.
Measurement of Tritium Released from Glucose as HTO. After storage, each 1-g sample was broken into
small fragments with a spatula. Two milliliters of distilled water was added, and the sample was resealed, equilibrated for 1 hr a t room temperature, frozen, and lyophilized. The distillate from each sample was collected in a separate trap with the lyophilization apparatus of Moss (1964). After lyophilization, the sample was broken into finer pieces, rehydrated with 2 ml of water, and lyophilized again. Radioactivity was measured in portions of the distillate by scintillation counting, and the lyophilization was repeated when tritium activity of the second distillate was greater than 10% of the first. Measurement of Color Formation. After lyophilizaEXPERIMENTAL SECTION tion, the color content of the sample was measured. The Materials. The casein and glucose were the same as in nonprotein-bound materials were separated from the prothe previous investigation (Clark and Tannenbaum, 1970). tein by washing them from the sample at the isoelectric ~ - G l u c o s e - 2 - ~ ~~C-,g l u c o s e - U - ~ ~~-glucose-6-t C, (Nupoint of casein (pH 4.6), followed by enzymatically solubiclear Chicago, Des Plaines, Ill.) and ~ - g l u c o s e - 6 - ~ ~ Clizing the protein with pronase (Clark and Tannenbaum, (New England Nuclear, Boston, Mass.) were supplied by 1970). Color content is equal to the absorbance a t 420 nm the manufacturer as 99% pure. times the sample volume; the total color included that Preparation of Systems. The casein-glucose model from the washings as well as the hydrolyzate. systems were prepared and stored as described previously RESULTS AND DISCUSSION (Clark and Tannenbaum, 1970) except for the addition of isotopically labeled sugars before freeze-drying. CarbonColor formation is presented in Figure 1 as a function of labeled glucose and glucose-6-t were added to each model time. After an induction period, the increase is linear. system studied to give specific activities of 9.0 or 4.5 pCi The rate drops off after about 20 days a t 55", suggesting T/mmol of glucose and 1.8 or 0.9 pCi 14C/mmol of glucose that some precursor becomes limiting at that point and for the 36 and 55" samples, respectively. The systems were the reaction is complete in about 100 days at 55". A t 36", designated CG-37d36-2C6-T, CG-37d36-6C6-T, CT-7d55the rate of color formation remains approximately conUC6-T, and C6-37d55-UC6-T. CG refers to casein-glucose, stant, increasing linearly with time (total brown color = 7d or 37d to the storage time in days, and 36 or 55 to the 283 ml). Whether this rate would eventually reach the temperature of storage a t 75% relative humidity. The last same plateau (500 ml expected a t approximately 400 days term in the code refers to the combination of isotopic lastorage) was not determined. bels used, such as 2-14C glucose and 6-t glucose for (-2C6The release of tritium, as HTO, from glucose-6-t is represented in Figure 2. The reactions involved in the tritium TI. Pigment Purification. Pigments were isolated and purelease are limited to about 12% of the total glucose-6-t. rified as previously described (Clark and Tannenbaum, Since brown color formation and HTO are similarly de1970). The procedures include isoelectric washing of the pendent on time, these data are plotted against each other protein, hydrolysis with proteolytic enzymes, and purifi(Figure 3). We performed a regression analysis (Kreyszig, cation by gel permeation chromatography on a 5 X 92 cm 1967) to determine the best-fitting straight line. column of Bio Gel P-4 (Bio-Rad Laboratories, Richmond, total brown color = 45.8 (T released) - 48.6 (1) Calif.) in 0.05 M NH40Ac buffer, pH 7.0. A portion of each fraction from the column was reserved for liquid The correlation coefficient is 0.991, indicating a relationscintillation counting, using a dioxane solvent system conship between brown color formation and HTO formation taining added water (Rapkin, 1967). The specific activity ( p < 0.001; Davies, 1961). The 14C and T specific activities of the eluates from the gel filtration column are presented in Figures 4 and 5. The ratios of these specific activities, normalized to that of the starting material, are plotted in Figure 6. The elution Department of Nutrition and Food Science, Massachucurve for the most severely stored sample (CG-37d55setts Institute of Technology, Cambridge, Massachusetts is very different from the others, apparently due to UC6-T) 02139. tritium loss from the C-6. Assuming that all of the C/T Present address: Corporate Research Department, The Coca-Cola Company, Atlanta, Georgia 30301. values for the lightly colored, less-severely stored samples 40
J. Agr. FoodChem., Vol. 21, No. 1, 1973
LIMIT-PEPTIDE PIGMENTS
- 500 E
w* 400
-
55 ‘C
a
2
bd
36 ‘C o---O
300
0 V
$
200
0 U
m
U
100
5
IU
’
O L
400
I
800
1600
1200
2000
IO
O
/
20
,4 0
DAYS
ELUTION
, OF
60
100
80
STORAGE
Figure 1. Formation of brown color in model system. Caseinglucose, 2: 1. 75% relative humidity.
20
40
OF
DbYS
60 STOR4GE
Figure 4. Elution profiles on BioGel P-4 (5
1200
800
400
ELUTION
500
X 92
cm, NH40Ac
pH 7.0) of radioactivity from labeled glucose with casein. 14C specific activity. Numbering code explained in text. CG-37d362C6T, 0 - - - - 0 ; CG-37d36-6C6T, 0 - - - - 0 ;CG-7d55-UC6-T, V - - - V ;CG-37d55-UC6-T, 0 - - - - 0 .
I00
80
Figure 2. Release of tritium from glucose-6-t with time of storage; system and storage conditions identical to Figure 1 .
--
I mi I
VOLUME
1600
VOLUME
2000
imll
Figure 5. Elution profiles of radioactivity from labeled glucose reacted with casein. Tritium specific activity (coiumn conditions and legend identical to Figure 4 ) .
-
E
2
-*
400-
4
300 -
0
U
A
37 * C 55 *C
J V
$
200
-
U
m ~
100-
U
k
0
I -0
2
4
6
8
10
12
“3:
?C3
TRITIUM
RELEASED,
%
OF
ADDED
LABEL
Figure 3. Correlation of tritium release from glucose-6-t with brown color formation. are the same, the mean and standard deviation of these data can be calculated (Davies, 1961). Most of the data for the dark, higher molecular weight pigments from CG37d55-UC6-T lie outside three standard deviations of this mean (Figure 6), which indicates a very significant difference between the samples. This difference is confirmed by the analysis of counting statistics (Herberg, 1964). The extent of the tritium loss is calculated from Figure 6, where the C / T ratio for the highest molecular weight fractions is 1.36 times that of the starting material. ( C / T sample)/(C/T starting sugar) = 1.36 (2) If we assume that the specific activity of the glucosederived carbon in the pigment is the same as in the start-
31
z3cn
8ECC
2Ul EL,.
.CLLUE
# _ I
’
Figure 6. Ratios of specific activities of radioactive iabels from starting material and reacted system in relation to elution profile (column conditions and legend identical to Figure 4 ) . ing sugar, then
T sample/T starting sugar = 0.74
(3)
which indicates that 26% of the tritium has been lost from pigment, compared to the carbon content. This loss represents a nearly threefold concentration of glucose molecules which have lost the 6-T in the pigment compared to only 9.3% 6-T lost from all the glucose present. Therefore, either the glucose that has lost 6-T is selectively incorporated into the pigment or glucose residues which are incorporated into the pigment selectively lose 6-T. The total loss of tritium from any fraction, as compared to the 14C content, is calculated from eq 4. J. Agr. Food
Chem., Vol. 21, No. 1 , 1973
41
CLARK, TANNENBAUM
- -
I
I
‘-OH II
A,-
c=o
-H20
C-OH
H20
I
C-OH
I
I1
I c=o I
c-0 I CH3
m
IT
I
“2.0 -
00
100 TOTAL
200
300
BROWN
COLOR
400
500
IN FRACTICN
I
mli
HC = 0 ( N H - R )
Figure 7. Correlation of release of tritium from glucose-6-f and brown color in high molecular weight ( > 2000 Daltons) purified pigments (sample CG-37d55-UC6-T, chromatography as in Figure 4).
H20
B.-
c
Y
total T loss = [fractional T loss] [sample weight (mg)]x I4C incorporation (dpm/mg)/(C/T) starting material (4)
I
(5)
A plot of the total tritium loss us. the total color (A420 volume) for each fraction (up to 1100 ml elution volume) is presented in Figure 7. In these fractions, where the tritium loss is large enough to be accurately measured, the tritium loss and the extent of color formation are highly correlated, which indicates that the purified pigments are truly representative of the overall system (Figure 3). In considering the tritium loss from glucose-6-t, one must also consider whether H or T or H and T are lost from the reaction involving the substituents on the C-6 and whether a kinetic isotope effect might be important. Since the immediate precursor for the tritium loss is unknown, this discussion is based on typical dehydration and enolization reactions which occur on glucose (Hodge, 1967; Figure 8). In example A, a n enediol (I) dehydrates to 11. Equilibration (enolization) between I1 and I11 in the presence of water would make all of the tritium on C-6 labile. Thus, a t equilibrium, any loss in T or H would result in loss of T. In example B, a degradation product, perhaps from hydrolytic ring opening of a 5-hydroxymethylfurfural residue (IV, linked or not linked to the peptide), would give V. V could enolize (V VI VI1 VIII), releasing only the T or H. In the presence of water, any enolization involving C-6 would result, a t equilibrium, in loss of T. VI11 might react with amines or condense with other sugar residues, which might or might not result in loss of T . Degradation reactions involving substituents on C-6 are thus likely to remove the tritium label. Therefore, the percentage of glucose molecules degraded a t C-6 can be assumed to be the same as the percentage which have released tritium. Tritium, with a molecular weight three times that of the common natural isomer, might react more slowly than H. Goldstein (1966) observed a large kinetic isotope effect when the rate-limiting step in a reaction sequence involved the breaking of a covalent bond of the heavy atom. Isbell et al. (1971) found a n isotope effect for enolization of sugars in alkaline solutions. The ratio of the rate constants for loss of T and H ( k T / k H ) was 0.13. It must be emphasized that storage time for this browning system was long compared to those for determination of enolization rates. We assume that enolic equilibrium has been reached and there is no isotope effect. It is important to note that only about one-fourth of the sugar residues in these highest molecular weight CGX
- - -
42
J. Agr.
Food Chem., Vol. 21, No. 1 , 1973
x
CHTOH
It I
yH2 (TIHO-C
where the fractional tritium loss is (from eq 3)
C-OH II CH I FI1H C-OH I
CHTOH
1 - (T sample/?‘ starting sugar) = fractional T loss
(NH-R)
I
700
6W
H20
I y 2 (TIHO-C
e
(H)T-C=O
4IIIT
II
(H)T-C-OH
ILU
--
Z0
I y 2 C=O I
CHTOH
PT
Figure 8. Possible reactions for loss of tritium from glucose-6-t residues of glucose-casein reaction.
37d55 pigments have lost 6-T. This represents, however, a threefold concentration effect compared to the 9.3% tritium released from the whole sample (Figure 2). It is unlikely that a pigment isolated from a sample stored for a longer time (-100 days) would have a much larger 6-T loss, since browning and 6-T loss are more than 80% complete at 37 days (Figures 1 and 2). If the loss of tritium from glucose is either the last slow step in the reaction sequence leading to pigment or if it occurs just after the last slow step, this could explain the high correlation shown in Figure 3. Alternatively, this loss of tritium could be unimportant; the reactions which form color could be competing for the same precursor(s). The tritium loss observed before color formation (Figure 3) could be related to these competing reactions or to a n isotope-exchange reaction. The high correlation of tritium loss and pigment molecular weight shown in Figure 7 indicates that tritium loss from C-6 is either essential for color formation and/or the cross-linkages of the pigment, or that it is a side reaction occurring shortly after the binding and crosslinking reaction(s). It is also possible, but less likely, that unbound sugars which have lost 6-T might be so unstable that they interact a t the site of pigment formation. Taken together, the correlations in Figures 3 and 7 are not conclusive, but strongly implicate terminal dehydration of the sugar molecule as a key crosslinking and chromophore-generating reaction. ACKNOWLEDGMENT
A. V. C. was supported by a Food Science Training Grant No. UI-01-033. The authors would like to thank John Hodge for helpful comments on the manuscript. LITERATURE CITED Chichester, C. O., S t a d t m a n , F. H., MacKinney, G., J. Amer. Chem. SOC.74, 3418 (1952). Clark, A. V.. Tannenbaum, S. R., J. Agr. Food Chem. 18, 891
SULFUR CHEMICALS
Hodge, J. E., “Syffiposiumon Foods: The Chemistry and Physiology of Flavors, Schultz, H. W., Day, E. A , , Libbey, L. M., Ed.. AVI Publ. Co.. Westuort. Conn.. 1967. DD 465-491. Isbell,’H. S., Linek,’K., Hepner, K. E., Jr:: ‘Carbohydr. Res. 19, 319 (1971). Kreyszig, E., “Advanced Engineering Mathematics,” Wilev, New York,-N. Y., 1967,p 822. Moss, G., J. Lab. C h . Med 63, 315 (1964).
Rapkin, E., “Laboratory Scintillation,” Vol 11, no. 31, Picker Nuclear, White Plains, N. Y., 1967, pp 1-10, Stadtman. F. H.. Chichester. C. 0.. MacKinnev. ., G.., J. Amer Chem. SOC. 74,3194 (1952). ’ Wolfrom, M. L., Schlicht, R. C., Langer, A. W., Jr., Rooney, C . S., J. Amer. Chem. SOC.75, 1013 (1953). ~
Received for review May 15,1972. Accepted September 27, 1972.
Isolation and Identification of Some Sulfur Chemicals Present in Two Model Systems Approximating Cooked Meat Cynthia J. Mussinan and Ira Katz* The following two model meat systems were heated: hydrolyzed vegetable protein (HVP)-Lcysteine.HC1-D-xylose-water and L-cysteineeHC1D-XylOS€’-Water.The flavor concentrates were isolated by atmospheric steam distillation, followed by continuous solvent extraction of the distillate. Isolation and identification were accomplished by One of the most challenging problems now confronting the flavor chem ,st is the successful identification and duplication of cooked meat flavor. The importance of this flavor type for the fabricated foods of today and of the future is clear to all those involved in developing protein foods for affluent as well as developing countries. During the past decade, numerous patents have been granted which encompass the processing of naturally occurring ingredients to produce a meat-like flavor. For instance, U. S. patent no. 3394015 (Giacino, 1968) concerns the processing of thiamine in the presence of cysteine and other amino acids to produce meat flavors. Also, U. S. patent no. 2934437 (Morton et al., 1960) is concerned with the processing of cysteine and other amino acids in the presence of pentose and hexose monosaccharides for the production of meat jlavors. Many workers, have speculated as to the importance of sulfur chemicals in meat flavor, but few such speculations have been reported. Brennan and Bernhard (1964) identified hydrogen sulfide and methyl, ethyl, propyl, and butyl mercaptans in the headspace of canned cooked beef. Dimethyl disulfide and dimethyl sulfone were identified by Liebich et a / (1972) in boiled beef. In addition, Chang e t al (1968) identified 3,5-dimethyl-1,2,4-trithiolan in boiled beef. It is our opinion that one of the major difficulties in the flavor analysis of cooked beef is that an important part of the meat flavor is due to trace constituents whose identification is made difficult by the large quantities of common lipid-derived aldehydes and ketones formed during the cooking process and present in the flavor isolate. For this reason, we decided to investigate initially a model system in order to give us some insight as to the types of sulfur chemicals present in a nonlipid system. EXPERIMENTAL SECTION Two model reaction systems were used throughout this work. In the first system, 3708 g of a carbohydrate-free
International Flavors & Fragrances, Inc., Union Beach, New Jersey 07735.
gas chromatography and coupled gc-mass spectrometry. Identifications were based on I E values and mass spectra. A total of 24 sulfur compounds were identified in the HVP-L-cysteine.HC1-D-xylose system and 15 were identified in the L-cysteine.HC1-D-xylose system, of which 10 were not present in the former model system. commercial hydrolyzed vegetable protein, 105.6 g of L-cysteine.HC1 (Diamalt A.G.), and 60.0 g of D-XylOSe (Fisher Scientific) were dissolvea in 8076 g of water, refluxed for 4 hr, and then allowed to stand overnight a t room temperature. The heated mixture possessed a strong roast odor reminiscent of cooked meat, coffee, and other roasted products. The heated product was atmospherically steam distilled in a 22-1. flask with a Kjeldahl bulb and an ice water cooled condenser. The distillate, 34 l., was collected a t a rate of 1 1. per hr in a 5” trap. The distillate was salt saturated and then continuously extracted for 7 hr with 700 ml of distilled diethyl ether in a liquid-liquid extractor. Acidic material was removed by extracting the ether extract with two 0.2-vol of 5% sodium carbonate. The ether extract was dried over anhydrous sodium sulfate and concentrated to approximately 10 ml by careful distillation in a Kuderna-Danish concentrator (Kontes Glass Co., Vineland, N. J.) equipped with a 300-mm x 13 mm i.d. Vigreux reflux column. The concentrate was divided into 17 fractions by small-scale preparative gas chromatography on a 20 ft X y4 in. 0.d. stainless steel column packed with 20% Carbowax 20M on 60-80 mesh AW-DMCS Chromosorb W. The instrument used was an F&M Model 700 gas chromatograph equipped with a thermal conductivity detector. The oven temperature was programmed from 70 to 225“ a t 2” per min after a 5-min post-injection hold. The temperature of the injector and detector was 230”, and the helium carrier gas flow rate was 80 ml per min. Repetitive 5O-pl injections were made and the effluent was collected in 150 mm X 2 mm 0.d. glass tubes cooled with crushed Dry Ice. The second model reaction system consisted of 36.3 g of D-xylose and 181.5 g of L-cysteine-HCl dissolved in 363 g of water. The solution was heated to 121” in a Parr bomb equipped with an agitator and internal cooling coil and held a t this temperature for 4 hr. After cooling, the contents were allowed to stand overnight a t 4”. The reaction product was extracted with two 300-ml vol of methylene chloride (Matheson, Coleman and Bell, ACS Reagent Grade), and the extract was dried and concentrated in a Kuderna-Danish concentrator as previously described. The concentrate was divided into ten fractions by smallin. scale preparative gas chromatography on a 10 ft x J. Agr. Food Chem., Vol. 21, No. 1, 1973
43
